{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:WCTASWGM3TATUY3OLIPYQZ2LGT","short_pith_number":"pith:WCTASWGM","schema_version":"1.0","canonical_sha256":"b0a60958ccdcc13a636e5a1f88674b34d4554f491c05dc184193862f4c7013fb","source":{"kind":"arxiv","id":"2307.02321","version":2},"attestation_state":"computed","paper":{"title":"MSViT: Dynamic Mixed-Scale Tokenization for Vision Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Amelie Royer, Babak Ehteshami Bejnordi, Jakob Drachmann Havtorn, Tijmen Blankevoort","submitted_at":"2023-07-05T14:22:31Z","abstract_excerpt":"The input tokens to Vision Transformers carry little semantic meaning as they are defined as regular equal-sized patches of the input image, regardless of its content. However, processing uniform background areas of an image should not necessitate as much compute as dense, cluttered areas. To address this issue, we propose a dynamic mixed-scale tokenization scheme for ViT, MSViT. Our method introduces a conditional gating mechanism that selects the optimal token scale for every image region, such that the number of tokens is dynamically determined per input. In addition, to enhance the conditi"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2307.02321","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2023-07-05T14:22:31Z","cross_cats_sorted":[],"title_canon_sha256":"5459ed002c08692309a33fdba29f03b09d5cefd1a1e37aac6e6558af84072834","abstract_canon_sha256":"2e3b6b579e5b312fe99ee6175bf9b612ebd197cd14da62e4aa32ce1a592a0ce1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:48:35.622979Z","signature_b64":"MxJnWLelgrF6RHa881Arjc9ihLo4wpKL/nE9AOpEPoQXQMpaTSVMpvspgVAyiQ98X0BjHDyljskk8dZIfL4SDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b0a60958ccdcc13a636e5a1f88674b34d4554f491c05dc184193862f4c7013fb","last_reissued_at":"2026-07-05T06:48:35.622516Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:48:35.622516Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MSViT: Dynamic Mixed-Scale Tokenization for Vision Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Amelie Royer, Babak Ehteshami Bejnordi, Jakob Drachmann Havtorn, Tijmen Blankevoort","submitted_at":"2023-07-05T14:22:31Z","abstract_excerpt":"The input tokens to Vision Transformers carry little semantic meaning as they are defined as regular equal-sized patches of the input image, regardless of its content. However, processing uniform background areas of an image should not necessitate as much compute as dense, cluttered areas. To address this issue, we propose a dynamic mixed-scale tokenization scheme for ViT, MSViT. Our method introduces a conditional gating mechanism that selects the optimal token scale for every image region, such that the number of tokens is dynamically determined per input. In addition, to enhance the conditi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2307.02321","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2307.02321/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2307.02321","created_at":"2026-07-05T06:48:35.622575+00:00"},{"alias_kind":"arxiv_version","alias_value":"2307.02321v2","created_at":"2026-07-05T06:48:35.622575+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2307.02321","created_at":"2026-07-05T06:48:35.622575+00:00"},{"alias_kind":"pith_short_12","alias_value":"WCTASWGM3TAT","created_at":"2026-07-05T06:48:35.622575+00:00"},{"alias_kind":"pith_short_16","alias_value":"WCTASWGM3TATUY3O","created_at":"2026-07-05T06:48:35.622575+00:00"},{"alias_kind":"pith_short_8","alias_value":"WCTASWGM","created_at":"2026-07-05T06:48:35.622575+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WCTASWGM3TATUY3OLIPYQZ2LGT","json":"https://pith.science/pith/WCTASWGM3TATUY3OLIPYQZ2LGT.json","graph_json":"https://pith.science/api/pith-number/WCTASWGM3TATUY3OLIPYQZ2LGT/graph.json","events_json":"https://pith.science/api/pith-number/WCTASWGM3TATUY3OLIPYQZ2LGT/events.json","paper":"https://pith.science/paper/WCTASWGM"},"agent_actions":{"view_html":"https://pith.science/pith/WCTASWGM3TATUY3OLIPYQZ2LGT","download_json":"https://pith.science/pith/WCTASWGM3TATUY3OLIPYQZ2LGT.json","view_paper":"https://pith.science/paper/WCTASWGM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2307.02321&json=true","fetch_graph":"https://pith.science/api/pith-number/WCTASWGM3TATUY3OLIPYQZ2LGT/graph.json","fetch_events":"https://pith.science/api/pith-number/WCTASWGM3TATUY3OLIPYQZ2LGT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WCTASWGM3TATUY3OLIPYQZ2LGT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WCTASWGM3TATUY3OLIPYQZ2LGT/action/storage_attestation","attest_author":"https://pith.science/pith/WCTASWGM3TATUY3OLIPYQZ2LGT/action/author_attestation","sign_citation":"https://pith.science/pith/WCTASWGM3TATUY3OLIPYQZ2LGT/action/citation_signature","submit_replication":"https://pith.science/pith/WCTASWGM3TATUY3OLIPYQZ2LGT/action/replication_record"}},"created_at":"2026-07-05T06:48:35.622575+00:00","updated_at":"2026-07-05T06:48:35.622575+00:00"}